摘要
Understanding soil water content, ¦È, variability is important for monitoring and modeling of land surface processes as well as land and water management practices. With regards to in situ ¦È probes, it is sometimes assumed that a single local measurement can represent the larger domain, mostly for practical reasons. But there is a substantial amount of variability in ¦È at the field scale. As part of the Bushland Evapotranspiration and Agricultural Remote Sensing Experiment 2008 (BEAREX08), a high-density sensor network and intensive observational periods were developed to fully describe the ¦È conditions at the surface on the field scale, in support of the hydro-meteorological measurements being collected. A total of 20 ¦È stations were distributed over an irrigated and a non-irrigated field (¡«10 ha each) and high-density (¡«every 5 cm) transects were measured for a high-detailed record. The network was able to provide large scale estimates of ¦È with an accuracy (root mean square error, RMSE) of 0.035 m3/m3. The network was temporally stable with the exception being immediately during and after irrigation events. Irrigation caused significant increases in coefficients of variation due to the length of time (8-12 h) necessary to irrigate the entire field. The spatial distribution of surface ¦È was significantly affected by the row structure of the cotton plants, which was North-South in the field where transect measurements were made with a row spacing of 76 cm. At scales <35 cm (approximately half the row spacing), the distribution was correlated in the East-West direction. For scales larger than 35 cm in the East-West direction the correlation decreased, but was still present. In the North-South direction this discontinuity was not present, and ¦È followed a power law distribution.